Skip to content

The Great Magenta Confusion

Observation card

Field Value
Status partial (teaching + sealed outcome plane; legacy HUD may still mislead)
Verified against 5ce15c13 docs; captures 2026-07-26
SceneId portal_gallery_seam_v0
SceneClass evidence_qualification
Channel ID (truth) cathedral.probe.outcome
Dependencies transport outcomes, not TextureRect readback
Units codes vs 8-bit display
Validity Do not select regions from RGB
Viz mapping NormalRGB + legacy magenta fill
Claim boundary Display ≠ probe
Evidence portal screenshots as presentation examples only

Retired wording

Retire

Magenta = no hit

This equation is false as doctrine. Magenta is a legacy fallback / presentation color whose meaning depends on the display path. It is not a ProbeOutcomeCode.

Authoritative: ProbeOutcomeCode or sealed channel cathedral.probe.outcome only.


Corrected public wording

RGB / NormalRGB / Depth / NdotV = Display Modes (presentation)
Legacy magenta blocks = host fallback fill — not ProbeOutcomeCode
Unresolved-budget = MaxStepsExhausted on the outcome plane
Background = BackgroundResolved (completed non-surface)
Surface = HitGeometry

Status

Item Status
Outcome plane as authority implemented
Live plate can show grid-aligned magenta observed
Legend “Magenta = no hit” on HUD legacy — retire from teaching
Outcome-driven hatch overlay partial / planned UI
RGB flood-fill region select forbidden

Screenshots (presentation examples only)

Portal A

File: screenshots/2026-07-26 Canonical Portal Scene with GRIN Film Camera Diagnostics normals.png

Element Caption role
3D room / portal world viewport
Film inset Observation Plate
Brown/blue fields display mapping (NormalRGB-like)
Rectangular magenta legacy presentation
Chrome / old legend legacy presentation (not validated semantics)

Safe observations: grid-aligned magenta (upscaled sample blocks); coarse orientation-like fields; possible glyph warp (transport presentation active); 80×45 + ~40% opacity reduce plate authority.

Not established from color alone: MaxStepsExhausted, no-hit, Unresolved Region membership.

Portal B

File: screenshots/2026-07-26 Canonical Portal Scene with GRIN Film Camera Diagnostics normals_B.png

Same roles: viewport + Observation Plate + display mapping + legacy presentation. Second presentation example only.


What must not be concluded from RGB alone

Forbidden Why
Magenta = no collider hit Plane may be all hits / all background
Magenta = MaxStepsExhausted Requires outcome code
Missing seam = transport failed May be shading/res/background-resolved
White/blue = semantic mesh class NormalRGB encodes n, not object labels

Reconnaissance doctrine: rendered magenta is not a trustworthy source for Unresolved Region selection.


What changed / why (teaching)

Layer Changes with field? Authority
World mesh No Structure
Glyph warp on portal Yes (transport presentation) Display of simulated transport
Magenta rectangle May persist Not outcome truth
Outcome histogram Policy-dependent Authoritative

Interpretation boundary

Claim boundary

Never build Region Probe selection from plate colors. Never equate legacy magenta with a single semantic code without the plane. Deeper refine is more Transport Effort, not automatic geometry. Region Probe (Cathedral architecture) does not prove wormholes.